>
>
>if the data has some missing value, how can i do when i use the "ezfftf". the
function seems not to allow the missing data.
>thank you!
>
No fast fourier transform allows missing data.
[Same for spherical harmonics.]
Two approaches main be used:
[1] Use some interpolation method to estimate the value(s)
at the missing locations. Note: Be aware of how the
interpolated values may affect the results of the FFT.
[2] Since the real and imaginary coefficients or, alternatively,
the amplitudes and phases are basically
least squares estimates at specified
frequencies, develop a code to make the estimates.
This would esentially be a "slow" fourier transform.
You could look at:
Fourier Analysis of Time Series: An Introduction
Peter Bloomfield
John Wiley
He has ways of estimating the real/imaginary coef of data
at user specified frequencies. [Just ignore missing data.]
---- There are books that address: Spectral Analysis of Unevenly Sampled Data Good Luck Dennis Shea_______________________________________________ ncl-talk mailing list ncl-talk@ucar.edu http://mailman.ucar.edu/mailman/listinfo/ncl-talk
This archive was generated by hypermail 2b29 : Thu Apr 22 2004 - 07:40:18 MDT